Professional Documents
Culture Documents
Quantum
Mechanics
MPAGS
October 2020
Contents
5 Observable quantities 32
5.1 Box normalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.2 Computing observable quantities from amplitudes . . . . . . . . . . . . . . 35
5.3 Decay rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.4 Cross sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.5 Mandelstam variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
i
PP2 Relativistic Quantum Mechanics
1
PP2 Relativistic Quantum Mechanics
• and understand how to perform calculations based on Feynman rules (spin sums/trace
techniques etc).
I am indebted to Michal Kreps, whose lectures this course is based on. Much of the content
of the course can also be found in chapters 3, 4, 5 and 6 of
1.2 Introduction
Why do we need Relativistic Quantum mechanics? In high energy physics we study
processes that involve “very small things travelling very fast”. For each aspect we have a
well tested theory:
The unification of general relativity (GR) and quantum mechanics (QM) is beyond the
scope of the MPAGS courses. There are numerous mathematical (quantum theories of
gravity are non-renormalizable) and conceptual issues (to do with locailty and unitarity)
with combining GR and QM. Attempts to unify GR and QM include string theory and
loop quantum gravity.
~ = (a1 , a2 , a3 ) , A
A ~ 0 = (a01 , a02 , a03 ) . (1.1)
~ and A
The vectors A ~ 0 correspond to the same physical object (they are the same vector)
and have the same magnitude (length) in each of the coordinate systems
~ 2 = (a1 )2 + (a2 )2 + (a3 )2 = (a01 )2 + (a02 )2 + (a03 )2 .
|A| (1.2)
where ê1 , ê2 , ê3 and ê01 , ê02 , ê03 are basis vectors describing the coordinate system of S and
S 0 , respectively. The coordinates are related through a linear transformation (a rotation)
3
X
i0
a = Rji aj . (1.4)
j=1
where the index i refers to a component of F~ . What form does this take in S 0 ?
F i = mai (1.6)
3
X 3
X
Rij F i = m Rij ai ,
i=1 i=1
0j 0j
F = ma
The form invariance comes from having objects that transform in the same way on the
left- and right-hand side of the equation.
x01 = x1 − vt
x02 = x2
(1.7)
x03 = x3
t0 =t
Newtons law’s are invariant under this type of transformation, but Maxwell’s equations
are not. Under a Galilei transformation,
dx0 dx
= −v ⇒ c0 = c − v (1.8)
dt0 dt
i.e. the speed of light is different for the different observers.
• the laws of physics are invariant (the same) for all inertial systems (systems of
coordinates moving with uniform velocity with respect to each other);
• the speed of light in the vacuum is the same for all observers.
where
v 1
β= , γ=p (1.10)
c 1 − β2
xµ = (x0 , x1 , x2 , x3 ) (1.11)
At this point we will introduce Einstein summation notation (contraction of indices) and
drop the explicit summation in the expression, i.e.
4
X
Λµν xν ≡ Λµν xν , (1.13)
ν=1
where a repeated pair of upper and lower indices implies a sum over that index such that
As in the case of 3-vectors we can construct lengths that are invariant under Lorentz
transformations, in this case the invariant length is
for example
we can see that the basis vectors for µ = ν = 1, 2, 3 are not orthogonal. Instead êµ · êµ = −1.
At this point it is useful to introduce the (metric) tensor gµν and a new type of vector
xµ = (x0 , −~x). We call this type of vector a co-vector. The metric tensor in Minkowski
space is defined by
such that
and similarly for g µν (which follows from gµν g µν = 1). The vectors and co-vectors are
related through the metric tensor,
xµ = gµν xν , (1.21)
which acts to lower and raise the indices. Anywhere we have a repeated upper and lower
index we have a dot product and get a scalar quantity, i.e. the dot product of A · B can
then be written as
Aµ Bµ = Aµ B µ = gµν Aµ B ν . (1.22)
i.e. the co-vectors transform with the inverse of the Lorentz transformation.
We can also write down the derivatives
µ ∂ ~ ∂ ~
∂ = , −∇ , ∂µ = ,∇ (1.25)
∂t ∂t
∂φ ∂xν ∂φ
∂µ0 φ = =
∂x0µ ∂x0µ ∂xν (1.26)
= [Λ−1 ]νµ ∂ν φ
i.e. M is the inverse of the Lorentz transform, M = [Λ−1 ]. The basis vectors of Minkowski
space transform as
We call Aµ a contravariant component and ~eµ a covariant basis vector. For the co-vectors,
the components are covariant and the basis vectors contravariant ,
9
PP2 Relativistic Quantum Mechanics
~ ·E
∇ ~ =ρ (Gauss’ law) ,
~
∇ ~ = − ∂B
~ ×E (Faraday-Lenz) ,
∂t (2.1)
~ ~
∇·B =0 (no magnetic charges) ,
~
∇ ~ = ~j + ∂ E
~ ×B (Ampere’s law) .
∂t
It turns out that these equations are Lorentz covariant, but this not immediately obvious
in this form. To Maxwell’s equations in a manifestly Lorentz covariant way, we start with
Ampere’s law and take its divergence
~
~ · (∇
∇ ~ × B)
~ =∇ ~ · ∂E
~ · ~j + ∇ (2.2)
∂t
∂
~ · ~j + (∇ ~
~ · E)
0=∇ (2.3)
∂t
~ · ~j + ∂ρ .
=∇ (2.4)
∂t
This is the continuity equation, which we can express in terms of 4-vectors
∂ ~
∂µ = , ∇ , j µ = (ρ, ~j) , (2.5)
∂t
such that
∂µ j µ = 0 . (2.6)
is locally conserved, i.e. the change in the charge in an arbitrarily small volume is only
due to the current flow out of the volume.
~ such that
At this point it is useful to introduce potentials, V and A,
~
~ =∇
B ~ ×A
~ , ~ = − ∂ A − ∇V
E ~ . (2.8)
∂t
~ ·B
By writing the magnetic and electric field in this way, the Faraday-Lenz law and ∇ ~ =0
~
are trivially satisfied. Re-writing Ampere’s law to include A and V
~ ×B
∇ ~ =∇
~ × (∇
~ × A)
~
!
∂ ~
∂A
= ~j + − ~
− ∇V (2.9)
∂t ∂t
~ ∇
= ∇( ~ · A)
~ −∇
~ 2A
~,
~0 = A
~ − ∇χ , ∂χ
A V0 =V + , (2.12)
∂t
where χ is a scalar field, this does not change the magnetic or electric fields. We can
choose χ to simplify our expressions, by picking a solution that sets
∇ ~ + ∂V = 0 .
~ ·A (2.13)
∂t
~ then the Lorentz gauge is
This is the so-called Lorenz gauge. If we write, Aµ = (V, A),
manifestly Lorentz invariant,
∂µ A µ = 0 (2.14)
Back to
2
∂ ~ ~ ~ 2V = ρ ⇒ ∂ V − ∇
~ 2V = ρ
− (∇ · A) − ∇ (2.15)
∂t ∂t2
and
2~
~ ∇
∇( ~ · A)
~ −∇ ~ = ~j − ∂ A − ∇
~ 2A ~ ∂V (2.16)
∂t 2 ∂t
which gives
~
∂ 2A ~ 2A
−∇ ~ = ~j . (2.17)
∂t 2
∂ µ ∂µ A ν = j ν (2.18)
where
∂2 ~2 .
∂µ ∂ µ = 2
−∇ (2.19)
∂t
This expression is also manifestly Lorentz covariant. You will sometimes see this written
with the replacement
≡ ∂µ ∂ µ so Aµ = j µ . (2.20)
∂µ j µ = 0 , ∂ µ ∂µ A ν = j ν (2.21)
with
• The eigenvectors form a complete set that spans the space such that
X
|ψm ihψm | = 1 (2.26)
m
Z
d3~x|~xih~x| = 1 (2.27)
pµ = i∂ µ . (2.33)
The operators Ê and p̂ operate on a wave-function ψ(~x, t) and this yields the Schrödinger
equation
∂ ~2 ~ 2
i~ ψ(~x, t) = − ∇ ψ(~x, t) + V (~x, t)ψ(~x, t) , (2.34)
∂t 2m
where V (~x, t) is an arbitrary potential.
We can also define a particle density by
ρ = ψ∗ψ (2.35)
Suppose the Klein-Gordon equation describes a free spin-0 particle of charge +e, e.g. a
π + . Then the electromagnetic-current associated with the charged particle is
17
PP2 Relativistic Quantum Mechanics
The wave-functions φ form a complete basis so the solutions to this equation can be
expressed as
X
ψ= am (t)φ(~x)e−iEm t (3.5)
m
where |am |2 is the probability to find the particle in state φm . This is time-independent
for a free system.
After we have introduced the potential
∂ψ
[H0 + V (~x, t)] ψ = i (3.8)
∂t
∂ X
=i am (t)φ(~x)e−iEm t ,
∂t m
the am (t) are time-dependent and we can get transitions between states. This can be
expanded as
X
[H0 + V (~x, t)] am (t)φ(~x)e−iEm t =
m
X ∂am (t) (3.9)
i − iEm am (t) φ(~x)e−iEm t
m
∂t
giving
" #
X ∂am (t) X
i φ(~x)e−iEm t = V am (t)φ(~x)e−iEm t (3.10)
m
∂t m
We can simplify this expression by multiplying through by φ∗f (~x) and integrating over the
phasespace, i.e.
Z X
∂am (t) ∗
i φf (~x)e+iEf t φm (~x)e−iEm t d3~x
m
∂t
Z X (3.12)
∗ +iEf t −iEm t 3
= am (t)φf (~x)e V (t, ~x)φ(~x)e d ~x
m
It is not possible to solve the system exactly, but we can exploit the weakness of the
potential and choose to pull out a small scalar coupling constant κ, i.e. to replace V (t, ~x)
by κV (t, ~x) and Vf m by κVf m . We can also express the am (t) as a power series in terms of
this small coupling constant as
X
am (t) = (κ)j ajm (t) , (3.16)
j
where
da0m
=0. (3.17)
dt
After comparing powers of κ, we have
daj+1
f (t)
X
= −i Vf m ei(Ef −Em )t ajm (t) (3.18)
dt m
At zeroth order
da0f (t)
= 0 ⇒ af = δf i (3.19)
dt
At second order
da2f (t) t
Z
0
X
= −i dt0 Vf m ei(Ef −Em )t a1m (t0 ) (3.23)
dt m −∞
Zt Z t0
i(Ef −Em )t0 0 00
a2f (t) = (−i)2
Vf m e dt Vmi ei(Em −Ei )t dt00 . (3.24)
−∞
−∞
where we have now substituted in the first order result. This process can then be repeated
for higher-and-higher orders in κ. It is also important to note that this process is time-
ordered.
If the potential is time-independent, i.e. V = V (~x), then
X 1
a2f (t) = −2πiδ(Ef − Ei ) Vf m Vmi , (3.25)
m6=i
Ei − Em + iε
which gives
Z
daf
= −i φ∗f (~x)V φi (~x)ei(Ef −Ei )t d3~x (3.27)
dt
and at time T /2 after the interaction has ceased,
ZT /2
Tf i ≡ af (T /2) = −i φ∗f (~x)V φi (~x)ei(Ef −Ei )t d3~xdt (3.28)
−T /2
We would like to interpret |Tf i |2 as the probability that a particle is scattered from state i
into state f . For the time-independent potential
Z T /2
Tf i = −iVf i ei(Ef −Ei )t dt (3.30)
−T /2
such that
sin2 ( T2 (Ef − Ei ))
|Tf i |2 = 4|Vf i |2 (3.31)
(Ef − Ei )2
This function is tightly peaked around Ef = Ei . If we take T → ∞
the δ-function expresses energy conservation in the transition. Typically we want to know
the transition rate from a known initial state to a group of final states (e.g. a particles
decay width),
X Z Z
2
|Ttot,f i | = |Tf i | = |Tf i | dN (Ef ) = |Tf i |2 ρ(Ef )dEf
2 2
(3.33)
f
sin2 ((Ef − Ei ) T2 )
Z
2 2
|Ttot,f i | = 4ρ(Ef )|Vf i | dEf (3.34)
(Ef − Ei )2
T appears on the right-hand side so this is not a transition rate. We can instead define a
transition rate per unit time as
|Ttot,f i |2
W = lim = 2π|Vf i |2 ρ(Ef ) (3.36)
T →∞ T
This is Femi’s golden rule. We made a lot of approximations to get here, but as long as the
interaction is weak this rule holds. In the QFT course, you will re-visit time-dependent
perturbation theory using operators and the Dirac interaction picture.
Coulomb scattering of
charged spin-0 particles
24
PP2 Relativistic Quantum Mechanics
(∂ µ ∂µ + m2 )φ = 0 . (4.1)
In classical EM, the motion of a particle with charge (−e) in a field Aµ can be obtained
by substituting
pµ → pµ + eAµ (4.2)
i∂ µ → i∂ µ + eAµ (4.3)
where ie∂ µ Aµ acts on φ, i.e. ie∂ µ (Aµ φ). This can be written in the form of an equation of
motion that depends on a potential
(∂ µ ∂µ + m2 )φ = −V φ , (4.5)
where
V = −ie(∂ µ Aµ + Aµ ∂µ ) − e2 Aµ Aµ . (4.6)
This is the interaction potential for the charged particle in the field. Since e 1,
e2 1
αEM = ∼ , (4.7)
4π 137
we can simplify the expression by taking only the first order terms
V ≈ −ie(∂ µ Aµ + Aµ ∂µ ) (4.8)
Assuming Aµ → 0 as xµ → ∞
+∞
φ∗f Aµ φi →0 (4.11)
−∞
Giving
Z
Tf i = −i (−ie)(φ∗f ∂µ φi − φi ∂µ φ∗f )Aµ d4 x (4.12)
(4.13)
This should look familiar, it is the Klein-Gordon current, only having different states for
φi and φf . From this we can define a transition current
and write
Z
Tf i = −i jµf i Aµ d4 x (4.15)
(4.16)
Now we can look at the initial and final states. These are both free spin-0 particles,
which obey the Klein-Gordon equation with solutions
φ = N e−ip·x (4.17)
Using
∂µ φ = −ipµ φ (4.18)
The final ingredient we need is the field Aµ . For coulomb scattering the source of this field
is the other particle. We already know that there will be a transition current associated
(2)
with this particle, jµ , and can use Maxwell’s equations to find an expression for the field,
µ
∂ µ ∂µ Aµ = j(2) = −ie(φ∗D ∂ µ φB − φB ∂ µ φ∗D ) (4.20)
= −e(pB + pD )µ NB ND ei(pD −pB )·x
We often write pD − pB = q.
Figure 4.2: Two particles scattering off each other. The particle four-momentum are indicated
by pA , pB , pC and pD . For coulomb scattering the internal line corresponds to the exchange of a
photon.
By realising that
gives
1
Tf i = −ie2 NA NB NC ND (pA + pC )µ 2 (pB + pD )µ
q
Z (4.26)
× ei(pC −pA +pD −pB )·x d4 x
which ensures 4-momentum conservation. We can also pull out an Lorentz invariant
amplitude
ig µν
−iM = ie(pA + pC )µ − 2 ie(pB + pD )ν (4.28)
q
It is interesting to note that in this process q 2 =
6 0, which implies that mγ 6= 0. The photon
is a “virtual” or “off-mass-shell” particle.
External photons are accompanied by a polarisation vector ε∗µ (p, λ), where p is the
photon momentum and λ is the spin projection of the photon along p~. We won’t derive the
origin of this factor in these notes. There are also two different internal lines depending
on whether a photon or a spin-0 particle are exchanged. We refer to these internal lines as
propagators.
The term
1
(4.31)
Ei − Em
in the second order expression looks a lot like a propagator. To make the connection to
the propagator its important to realise that when we draw a Feynman diagram we really
mean a sum over different possible time-ordered diagrams. When we do the calculation
we integrate over space-time and enforce energy conservation for the external particles.
Figure 4.4: Feynman diagrams are a sum over the different possible time-ordered diagrams.
For the first time ordering in Fig. 4.4, the energy of the intermediate state is
p
Em = EC + EB + |~pA − p~C |2 + m2 , (4.32)
p
where the exchanged particle has energy EP = |~pA − p~C |2 + m2 and mass m. The energy
of the initial state is
Ei = EA + EB . (4.33)
This gives
p
Ei − Em = (EA + EB ) − EC − EB − |~pA − p~C |2 + m2 (4.34)
p
= (EA − EC ) − |~pA − p~C |2 + m2 .
The same exercise for the second time ordering in Fig. 4.4 gives
p
Em = EA + ED + |~pB − p~D |2 + m2 (4.35)
and
p
Ei − Em = −(EA − EC ) − |~pA − p~C |2 + m2 . (4.36)
Observable quantities
32
PP2 Relativistic Quantum Mechanics
φ = N e−ip·x , (5.1)
j µ = 2|N |2 pµ (5.2)
ρ = j 0 = 2E|N |2 . (5.3)
dn = ρ(E)dE . (5.5)
Figure 5.1: Box normalistion of states. States exist in box with sidelength L.
If the states exist in a box of side L, we know from the infinite square well problem in
Quantum Mechanics that in one dimension (x),
2π
px = nx , (5.6)
L
where nx is an integer. This gives
2π
dpx = dnx . (5.7)
L
The total number of states in a cube of side L with momentum between p~ and p~ + d3 p~ is
then
Here, pf and pi are the final and initial state 4-momentum summing over all of the particles.
The Nf and Ni are the normalisations associated to each of the initial and final state
particles. The δ-function ensures energy and momentum conservation in the process.
Figure 5.2: Illustration of a typical scattering problem in High Energy Physics, occuring in a
local volume V over time T .
We can turn this amplitude into a probability for a transition from i → f by taking its
square,
2
Y Y
pf i = |Tf i |2 = − iMδ 4 (pf − pi ) Nf Ni . (5.11)
f i
This includes the square of a δ-function. To deal with this can replace the δ-function with
the identity
Z
(2π) δ (pf − pi ) = ei(pf −pi )·x d4 x ,
4 4
(5.12)
such that
Z
4 4 2 4 4
|(2π) δ (pf − pi )| ≈ (2π) δ (pf − pi ) ei(pf −pi )·x d4 x (5.13)
T,V
where the integral is over the volume of the interaction region, V , and the extent of the
interaction in time, T . The first δ-function forces pf = pi , so this is simply
|(2π)4 δ 4 (pf − pi )|2 ≈ V T (2π)4 δ 4 (pf − pi ) (5.14)
Putting the ingredients back into |Tf i |2 ,
Y Y
|Tf i |2 = |M|2 V T (2π)4 δ 4 (pf − pi ) |Nf |2 |Ni |2
f i (5.15)
1 1
= |M|2 V T (2π)4 δ 4 (pf − pi ) nf ni
V V
where nf and ni are the number of initial and final state particles. For a process involving
A + B → C + D, we would have
|Tf i |2 = |M|2 V T (2π)4 δ 4 (pC + pD − pA − pB )|NA |2 |NB |2 |NC |2 |ND |2 (5.16)
1
= |M|2 V T (2π)4 δ 4 (pC + pD − pA − pB ) 4
V
The transition rate per unit volume can be computed from |Tf i |2 as
|Tf i |2 1 1
Wf i = = |M|2 (2π)4 δ 4 (pf − pi ) nf ni . (5.17)
VT V V
The transition rate per unit volume into a small fixed final state phase-space is
dW = Wf i dn (5.18)
1 1 Y d3 p~
= |M|2 ni fi (2π)4 δ 4 (pf − pi ) V
V V f
2Ef (2π)3
1 Y d3 p~
= |M|2 (2π)4 4
δ (p f − p i )
V ni f
2Ef (2π)3
(5.19)
We call
Y d3 p~
dQ = (2π)4 δ 4 (pf − pi ) (5.20)
f
2Ef (2π)3
rearranging gives
EB EC
|~p|d|~p| = d(EB + EC ) . (5.31)
EB + EC
We can use this to write
d3 p~ |~p|2 d|~p|dΩ |~p|d(EB + EC )dΩ
= = , (5.32)
EB EC EB EC EB + EC
and get
|~p|d(EB + EC )dΩ
Z
1
Γ= |M|2 δ(mA − EB − EC ) (5.33)
8(2π)2 mA EB + EC
2 4 4 |M|2
Wf i = |Tf i | = (2π) δ (pC + pD − pA − pD ) 4 (5.39)
V
This gives
1 1 4 d3 p~C d3 p~D
dσ = |M|2 δ (p C + p D − p A − p D ) , (5.42)
4EA EB |~vA | (2π)2 2EC 2ED
where once again the arbitrary normalisation volume cancels. It’s useful to write this in
terms of the Lorentz invariant phase-space
|M|2
dσ = dQ , (5.43)
F
where F is the incident flux
dσ 1 |~pC |
= 2 2
|M|2 . (5.46)
dΩ 64π (pA + pB ) |~pA |
s = (pA + pB )2 = (pC + pD )2
t = (pA − pC )2 = (pB − pD )2 (5.47)
u = (pA − pD )2 = (pB − pC )2
These are called Mandelstam variables. They will appear in scattering calculations and
are Lorentz covariant, they can be used to describe a process in a frame independent way.
Calculating cross-sections
for spin-0 particle scattering
40
PP2 Relativistic Quantum Mechanics
2. Using the Feynman rules, label the elements in each diagram with the appropriate
factor.
3. Form an invariant amplitude −iM by multiplying all of the factors in the proper
order. This is the order which follows the lines in the diagram against the direction
of the particle flow.
where we have used q = pA − pC . If there were more diagrams, we would sum over them
in the amplitude. The transition probability
2
(pA + pC )µ (pB + pD )µ
2 4
|Me− µ− | = e (6.2)
(pA − pC )2
2
4 pA · pB + pA · pD + p C · pB + pC · pD
=e
t
s = (pA + pB )2 = (pC + pD )2
= p2A + p2B + 2pA · pB = m2A + m2B + 2pA · pB
= p2C + p2B + 2pC · pD = m2C + m2D + 2pC · pD
(6.3)
2 2
u = (pA − pD ) = (pB − pC )
= p2A + p2D − 2pA · pD
= p2B + p2C − 2pB · pC ,
as
pA · pB + pA · pD + pC · pB + pC · pD = (6.4)
1
2
(s − m2A − m2B ) + 12 (s − m2C − m2D ) − 12 (u − m2A − m2D ) − 12 (u − m2B − m2C )
such that
2
2 4 s−u
|Me− µ− | = e .
t
In the centre of mass system |~pA | = |~pC |. Further, in the very high energy limit me Ee
and mµ Eµ and the differential cross-section becomes
2
e4
dσ 3 + cos θ
= , (6.6)
dΩ 64π 2 s 1 − cos θ
CM
where, θ is the angle between p~A and p~C . The cross-section diverges as θ → 0.
Figure 6.3: Diagrams for e− e− → e− e− scattering. Note, in the second diagram the two lines do
not cross.
It’s easy to see that this is invariant under C → D. It’s also invariant under B → D. This
In order to write down the invariant amplitude we need Feynman rules involving an-
tiparticles. If we recall the Feynman-Stückelberg picture then a positive energy antiparticle
travelling forwards in time is equivalent to a negative energy particle travelling backwards
in time.
We can use the Feynamn-Stückelberg picture for example to write the vertex term
for the lower vertex in the first diagram as ie(−pB − pD )ν . For the first diagram, the
4-momentum of the exchanged photon as q = pA − pC . For the second diagram, the
4-momentum of the photon is q = pA + pB . From the diagrams, the invariant amplitude is
then
µ µ
2 (pA + pC )µ (−pB − pD ) 2 (pA − pB )µ (pC − pD )
−iMe e = ie
+ − + ie (6.9)
(pA − pC )2 (pA + pB )2
Comparing this to the expression we had earlier we find that
Me− e− (pA , pB , pC , pD ) = Me+ e− (pA , −pD , pC , −pB ) (6.10)
i.e. it is the same expression just exchanging pB ↔ −pD . We can exploit this to short cut
to the result
s = (pA + pB )2 → u
t = (pA − pC )2 → t (6.11)
u = (pA − pD )2 → s
such that
2 u−s u−t
−iMe+ e− = ie + (6.12)
t s
We refer to the diagram that contributes (u − s)/t to the amplitude as a t-channel process
and the diagram that contributes (u − t)/s as an s-channel process.
47
PP2 Relativistic Quantum Mechanics
The motivation for this choice will become clear later. The Pauli matrices obey the
commutation relation
[σ i , σ j ] = 2iεijk σ k (7.12)
where
0 if i, j, k repeat
εijk = +1 for cyclic permutations (7.13)
−1 for non-cyclic permutations
with γ µ = (β, βαi ). Explicitly writing out the sum over the space-time indices
∂ψ ∂ψ ∂ψ ∂ψ
iγ 0 + iγ 1 1 + iγ 2 2 + iγ 3 3 − m14 ψ = 0 . (7.16)
∂t ∂x ∂x ∂x
The γ matrices have the following useful properties that we’ll need in our later calculations
γ µ γ ν + γ ν γ µ = 2g µν 1
γ 0† = γ 0
γ 0 γ 0 = +14
(7.18)
γ i† = −γ i
γ i γ i = −14
γ µ† = γ 0 γ µ γ 0
To illustrate where this expression comes from, we can take the Hermitian conjugate of
the Dirac equation
−i∂µ ψ † γ µ† − mψ † 1 = 0 . (7.20)
−i∂µ ψ † γ µ† γ 0 − mψ † γ 0 = 0 (7.21)
† 0 µ † 0
−i∂µ ψ γ γ − mψ γ = 0
(7.22)
i.e.
∂µ (ψ̄γ µ ψ) = ∂µ j µ = 0 . (7.24)
ρ = j 0 = ψ̄γ 0 ψ =ψ † γ 0 γ 0 ψ (7.25)
=ψ † ψ > 0 .
ψ = u(p)e−ip·x , (7.27)
where we call u(p) a spinor and e−ip·x is associated with the usual plane wave solution.
Substituting this solution into the Dirac equation yields
which collapses to
(γ µ pµ − m)u(p) = 0 (7.29)
To learn something about the spinor u(p) it is useful to express the γ matrices once again
in terms of α
~ and β,
α · p~ + βm)u(p) = E u(p) .
H u(p) = (~ (7.31)
There are four independent solutions to this equation, two with E > 0 and two with E < 0.
For a particle at rest
m12 0
β m u(p) = u(p) = E u(p) (7.32)
0 −m12
For p~ 6= 0,
m12 ~σ · p~ uA uA
=E (7.34)
~σ · p~ −m12 uB uB
where u(p) has been divided into two two-component spinor uA and uB . This can be
written more compactly as
(~σ · p~ ) uB = (E − m) uA (7.35)
(~σ · p~ ) uA = (E + m) uB (7.36)
(7.37)
where
p3 p1 − ip2
~σ · p~ = . (7.38)
p1 + ip2 −p3
Solutions only exist if
−(E − m)1 ~σ · p~
det =0 (7.39)
~σ · p~ −(E + m)1
and we again arrive at E 2 = |~p|2 + m2 , which has positive and negative energy solutions.
The two component spinors are, at this point, arbitrary i.e. the two component spinors
can be written as
uA = a1 χ1 + a2 χ2 (7.40)
uB = b1 χ2 + b2 χ2
with
1 1 2 0
χ = , χ = . (7.41)
0 1
For the E > 0 solutions, we can take uA (s) = χ(s) (where s = 1, 2) and
~σ · p~ (s)
uB (s) = χ (7.42)
E+m
The positive-energy four spinors solutions to the Dirac equation are then
χ(s)
u(p, s) = N σ ·~
~ p (7.43)
E+m
χ(s)
where N is a normalisation constant. For E < 0, we can instead write uB (s) = χ(s) and
solve for uA such that
~σ·~p (s) σ ·~
~ p
χ(s)
E−m
χ − |E|+m
u(p, s + 2) = N =N (7.44)
χ(s) χ(s)
The four solutions are orthogonal, such that u† (r)u(s) = 0 if r 6= s.
For the normalisation, we once again choose N such that we have 2E particles per unit
volume,
2 !
~σ · p
~
u† (p, s)u(p, s) = |N |2 1 + = 2E . (7.45)
E+m
53
PP2 Relativistic Quantum Mechanics
8.1 Spin
In addition to the positive and negative energy solutions we have an additional two-
fold degeneracy (which gives four eigenvectors/eigenvalues). To understand where this
degeneracy comes from, we need to find out if there is another observable that commutes
with the Hamiltonian.
The Hamiltonian for a free particle is
~ · p~ + βm
H0 = α (8.1)
~ = ~x × p~ commutes with H0 ,
First let’s check to see if angular momentum L
~ = [~
[H0 , L] α · p~ + βm, ~x × p~] (8.2)
α · p~, ~x × p~]
= [~ (8.3)
Using
gives
~ = −i~
[H0 , L] α × p~ (8.5)
i.e. angular momentum is not a good quantum number. The Dirac equation does not
conserve orbital angular momentum and therefore cannot describe a spin-0 particle. Now
lets consider another operator
~ = ~
σ 0
Σ (8.6)
0 ~σ
which should be familiar from your quantum mechanics classes.
~ = 2i~
[H0 , Σ] α × p~ (8.7)
[H0 , J] ~ + 1 Σ]
~ = [H0 , L ~ =0 (8.8)
2
The operator S~ = 1Σ~ is intrinsic to the particle. For a particle at rest S
~ commutes with
2
H0 and is a good quantum number, e.g. operating through with S3 = 12 Σ3
which is the projection of the spin along the particles momentum vector. We refer to this
as the particles Helicity. If we choose p~ = (0, 0, p)
1
E>0: Σ
2 3
u(E > 0, 1) = + 12 u(E > 0, 1)
1
Σ
2 3
u(E > 0, 2) = − 12 u(E > 0, 2)
(8.11)
1
E<0: Σ
2 3
u(E < 0, 3) = + 12 u(E < 0, 3)
1
Σ
2 3
u(E < 0, 4) = − 12 u(E < 0, 4)
Eigenvalues of + 12 have spin aligned with the particles momentum vector. We have found
~ the total angular
that the Dirac equation describes a spin- 12 particle and conserves, J,
momentum.
Figure 8.1: Helicity and handiness of a particle with a momentum vector pointing along the
direction p̂.
Asside:
Helicity is also not a very well defined quantity. We can always boost to the rest-
frame of the particle where the helicity is undefined. A more clearly defined observable
(relativistically covariant observable) is the particles chirality, related to the operator
γ 5 = iγ 0 γ 1 γ 2 γ 3 . (8.12)
In the Pauli-Dirac basis, which we are working with for the γ-matrices, this is
5 0 12
γ = . (8.13)
12 0
We can then define left- and right-handed projection operators, which for relativistic
particles project out the left- and right-handed helicities as
PL = 12 (1 − γ 5 )
(8.14)
PR = 12 (1 + γ 5 ) .
8.2 Antiparticles
It’s interesting to look at how Dirac interpreted the negative energy solutions. He postulated
the existence of a sea of negative energy states and that the vacuum (or ground state of
the system) has all of these negative states full. An additional electron must fill one of the
positive energy states because the Pauli exclusion principal forbids it from occupying a
filled negative energy states. If we promote a negative energy state to a positive one, an
electron-hole pair is created (a positive energy electron and a hole in the negative sea). The
hole appears as a positive energy anti-particle (the positron). This mechanism is essentially
particle and antiparticle pair production. The positron was only observed experimentally
four years later by Anderson in a cloud chamber experiment. Dirac’s mechanism only
works for fermions because bosons do not obey the Pauli exclusion principal.
We have solutions for the Dirac equation that can be written for the E > 0 solutions as
χs
−ip·x 1/2
ψ(p+ ) = u(p+ , s)e , u(p+ , s) = (E + m) σ ·~
~ p (8.15)
E+m
χs
where
If we think of the Feynman-Stückelberg prescription, then we also need to swap the sign
of p~ to interpret the negative energy solutions in terms of anti-particles i.e.
~σ·~p s
1/2 χ
ψ(p)anti = (E + m) E+m
s e+ip·x . (8.18)
χ
We can introduce a second type of spinor for the antiparticles, v(p, s). The v-spinors are
spinors for anti-particles with 4-momentum (E, p~) and we keep the u-spinors for particles
with E > 0. The v-spinors are also orthogonal,
χ1
−ip·x 1/2
ψ(p, 1) = u(p, 1)e = (E + m) σ ·~
~ p
E+m
χ1
(8.20)
χ2
−ip·x 1/2
ψ(p, 2) = u(p, 2)e = (E + m) σ ·~
~ p .
E+m
χ2
Note, the association of v(p, 1) with u(p, 4) and v(p, 2) with u(p, 3). This is due to the
change in the momentum direction and ensures that the indices 1 and 2 in ψ(p, s) refer to
the positive and negative helicity states, respectively.
The current associated to the Dirac equation is
j µ = ψ̄γ µ ψ (8.22)
or
We also need to define adjoint spinors for our Dirac current j µ = ψ̄γ µ ψ. To obtain an
adjoint spinor we can take the Hermittian conjugate,
†
(p/ − m)u(p, s) = ū(p, s)(p/ − m) = 0 (8.26)
and similarly
where
The combination
With the adjoint spinors we have a complete set of states and can write the completeness
relation, summing over the spins
X
u(p, s)ū(p, s) = p/ + m
s=1,2
X (8.32)
v(p, s)v̄(p, s) = p/ − m
s=1,2
i∂ µ → i∂ µ + eAµ , (8.33)
we get
There should also be an equivalent expression for a spin- 21 particle with a charge of +e
The form of the equation has to be the same because nature cannot care about how we
define the charges. An obvious question is, what is the relationship between ψ and ψC
(and can we define an operator that transforms between the two)? For the Klein-Gordon
equation, we simply took the complex conjugate. Taking the complex conjugate of the
Dirac equation gives
ψC = Ĉψ ∗ . (8.38)
−Ĉγ µ∗ = γ µ Ĉ , (8.39)
such that
It must also be Hermitian, i.e. Ĉ † Ĉ = 1. We discussed earlier that the γ-matrices are not
unique and the exact form of Ĉ will depend on the representation of the γ-matrices. In
the Pauli-Dirac representation it is
+1
2 0 iσ2 −1
Ĉ = iγ = = (8.41)
−iσ2 0 −1
+1
This is just
The operator Ĉ is the charge conjugation operator. It changes a particle into an anti-
particle and vice versa, flipping the sign of all of the particles charges.
Asside:
The choice of γ matrices is not unique. In this course we are working with the so-called
Pauli-Dirac basis,
0 σi 12 0
αi = , β= . (8.43)
σi 0 0 −12
The Pauli-Dirac basis disgonalises in the particle energy in the non-relativistic limit. This
leads to a particularly simple looking set of eigenvalues and eignevectors for a particle at
rest. The Weyl basis is not diagonalised in the particle energy (the unit matrices in β
appear off-diagonal) and instead diagonalises the particle helicity in the relativistic limit
(such that the unit matrices in γ 5 appear on the diagonal).
Coulomb scattering of
charged spin- 12 particles
63
PP2 Relativistic Quantum Mechanics
Once again we need to find the form of the potential, V (x). We can start by writing the
Dirac equation in a form that separates out the Hamiltonian. The Dirac equation is
iγ 0 ∂0 + iγ j ∇j + eγ µ Aµ − m ψ = 0
(9.7)
Figure 9.1: Comparison of Feynman rules for spin-0 and spin- 21 particles.
Comparing to the free particle case (with H = H0 + V ), we can make the association
V = −eγ 0 γ µ Aµ . (9.9)
Plugging the potential into the transition amplitude expression gives
Z
Tf i = −i ψf† (x)V (x)ψi (x)d4 x (9.10)
Z
= ie ψf† γ 0 γ µ Aµ ψi d4 x
Z
= ie ψ̄γ µ Aµ ψi d4 x
where jfµi was the transition current associated to the other particle. The dependence on
1/q 2 came from the realisation that
1 −iq·x
∂ µ ∂µ e−iq·x = − e (9.14)
q2
where q = pi − pf .
Finally, we get the expression for the transition amplitude for A + B → C + D,
Z
(1) 1 µ
Tf i = −i jµ (x) − 2 j(2) (x)d4 x (9.15)
q
1
= −i(−eūC γµ uA ) − 2 (−eūD γ µ uB )(2π)4 δ 4 (pA + pB − pC − pD )
q
where q = pA − pC . In terms of the invariant amplitude,
where
µ −igµν
−iM = (ieūC γ uA ) (ieūD γ ν uB ) (9.17)
q2
All of the dynamics of the system is in the invariant amplitude, M.
At this point it is useful to look in more detail at the structure of the Dirac current,
ūf γ µ ui . To do this we make a so-called Gordon-Decomposition,
1
ūf γ µ ui = ūf ((pf + pi )µ + iσ µν (pf − pi )ν ) ui (9.18)
2m
where
σ µν = 2i [γ µ , γ ν ] = 2i (γ µ γ ν − γ ν γ µ ) (9.19)
The first term in the decomposition looks identical to what we had in the spin-0 case
for the vertex factor. The second term is a magnetic moment and is related to the
spin of the particle. This is new for the spin- 12 case. If you would like to prove the
Gordon-Decomposition works you can expand the right-hand side and then simplify using
These factors come from the anticommuation properties of the fermionic operators (and
appear when ordering the fermionic fields in a full Quantum Field Theory treatment).
This looks complicated to expand. Don’t worry, it’s not as complicated as it looks. The
structure of ūγ µ u is
For spin- 12 particles, there are two possible spin-states. We have already seen that summing
of the spins
X
u(p, s)ū(p, s) = (p/ + m) (9.33)
s
71
PP2 Relativistic Quantum Mechanics
γ 0 γ 0 = +1
γ i γ i = −1 (10.1)
{γ µ , γ ν } = γ µ γ ν + γ ν γ µ = 2g µν
γ 5 = iγ 0 γ 1 γ 2 γ 3 (10.2)
such that
γ 5 γ 5 = 14
γ 5† = γ 5 (10.3)
{γµ , γ5 } = γ µ γ 5 + γ 5 γ µ = 0 .
You will come across the γ 5 matrix again when you deal with interactions involving the
weak interaction which have both vector ūγ µ u and axialvector ūγ µ γ 5 u couplings.
The trace of an n × n matrix is
N
X
Tr(A) = Aii , (10.4)
i=1
where the Aii are the diagonal elements of the matrix. The 4 × 4 unit matrix has a trivial
trace
Tr(1) = 4 . (10.5)
Traces are also the same for cyclic permutations, such that
a/b) = 4a · b = 4aµ bµ
• Tr(/
• Tr(γ 5 ) = 0
• Tr(γ 5 a
//b) = 0
• Tr(γ 5 a
//b/cd/) = 4iεµνλσ aµ bν cλ dσ
Tr(γ α γ σ γ ρ . . . γ µ ) = Tr(γ α γ σ γ ρ . . . γ µ γ ν γ 5 γ 5 )
= Tr(γ 5 γ α γ σ γ ρ . . . γ µ γ ν γ 5 )
(10.7)
= Tr(γ 5 γ α γ σ γ ρ . . . γ µ γ 5 γ ν ) × (−1)
= Tr(γ 5 γ 5 γ α γ σ γ ρ . . . γ µ γ 5 γ ν ) × (−1)n
where n is the number of γ matrices and we have used the anti-commutation relation
{γ µ , γ 5 } = 0 and γ 5 γ 5 = 1. If n is odd the trace has to be zero for the left- and right-hand
sides to be equal.
a/b) = 4a · b
Example: To show that Tr(/
a/b) = Tr(/ba
Tr(/ /)
1
a/b + /ba
= 2 Tr(/ /)
= 12 aµ bν Tr(γ µ γ ν + γ ν γ µ )
(10.8)
= 12 aµ bν Tr(2g µν 1)
= g µν aµ bν Tr(1)
= 4g µν aµ bν = 4aν bν
Lµν 0 0 0 0 µ ν
(e) = Tr(u(k , s )ū(k , s )γ u(k, s)ū(k, s)γ ) (10.9)
1 e4 0 µ ν
0
|M|2 = Tr (k/ + m(e) )γ ( /
k + m(e) )γ × Tr ( p
/ + m(µ) )γµ (p/ + m(µ) )γν . (10.11)
4 q4
Both traces have the same structure, so we only need to solve it once,
1 0
Lµν µ ν
(e) = Tr ( /
k + m (e) )γ (k/ + m (e) )γ (10.12)
2
1 0 0
= Tr k/ γ µ k/γ ν + k/ γ µ m(e) γ ν + m(e) γ µ k/γ ν + m2e γ µ γ ν .
2
The easiest term to handle is
Tr (p/γ µ mγ ν ) . (10.13)
This contains an odd number of γ matrices and so the trace is zero. It is also easy to
handle the term
m2
Tr (m2 γ µ γ ν ) = Tr(γ µ γ ν + γ ν γ µ ) (10.14)
2
m2
= Tr(2g µν 1) = 4m2 g µν . (10.15)
2
The remaining term is
Lµν 0µ ν 0ν µ µν 0 2
(e) = 2(k k + k k − g (k · k − m(e) )) . (10.17)
Finally,
e4 µν (µ)
|M|2 = L Lµν (10.18)
q 4 (e)
4e4
= 4 (k 0µ k ν + k 0ν k µ − g µν (k 0 · k − m2 ))(p0µ pν + p0ν pµ − gµν (p0 · p − m2(µ) )) . (10.19)
q
We now need to multiply all the pieces and keep track of the indices we are summing, e.g.
k µ k 0ν pµ pν = (k · p)(k 0 · p0 ) . (10.20)
8e4 0 0 0 0 2 0 2 0 2 2
|M|2 = (k · p)(k · p ) + (k · p )(p · k) − m(e) (p · p ) − m(µ) (k · k ) + 2m(e) m(µ)
q4
(10.21)
8e4
|M|2 = [(k · p)(k 0 · p0 ) + (k · p0 )(p0 · k)] , (10.22)
q4
and as before this can be expressed in a frame invariant form in terms of the Mandelstam
variables,
as
s2 + u2
4
|M|2 = 2e (10.24)
t2
pB ↔ −pC (10.25)
4πα2
σ(e− e+ → µ− µ+ ) = (10.29)
3s
This is the first result that we can compare to data (see Fig. 10.2). It agrees quite nicely,
but only in this energy range. At higher energies we can also draw diagrams involving the
Z boson that will have√ a significant impact on the cross-section (there will be a pole-like
enhancement when s = mZ ).
Figure 10.2: Total cross section for e− e+ → µ− µ+ measured at the PETRA accelerator at DESY.
The solid-curve corresponds to our QED expectation.
Aµ = 0 (10.31)
i.e. setting the current density to zero. This equation has solutions of the form
Aµ = εµ (~q)e−iq·x (10.32)
Substituting this back into Aµ = 0 gives q 2 = 0, i.e. mγ = 0 (as expected). The
four-vector εµ (~q) deals with the polarisation of the photon. The photon is a spin-1 particle,
which has only two transverse polarisation states and so we expect to be able to cancel
two of the four possible polarisation vectors. In general, a spin-1 particle with non-zero
mass will have three polarisation states. How do we remove two of the polarisation states?
The Lorenz condition ∂µ Aµ gives
qµ ε µ = 0 , (10.33)
reducing the number of independent components to three. We also have freedom in out
choice of gauge, transforming
Aµ → Aµ + ∂µ Λ . (10.34)
Choosing
Λ = iae−iq·x (10.35)
ε → ε + aqµ (10.36)
i.e. two polarisation vectors that differ by multiples of qµ describe the same photon. We
can ue this freedom to ensure the time component of the polaristion vanishes, i.e. ε0 = 0.
The Lorentz condition reduces to
~ε · ~q = 0 (10.37)
For a photon travelling along the z-axis, the two polarisation vectors are
[2]
[2]
~j = 1 ψ ∗ ∇ψ
~ − ψ ∇ψ~ ∗ .
2mi
[4]
[2]
[3]
3. In the very high-energy limit (E m), show that the differential cross section for
spinless electron-muon scattering in the centre of mass system becomes
2
α2 3 + cos θ
dσ
= ,
dΩ 4s 1 − cos θ
CM
2
where α = e /4π and θ is the scattering angle.
[3]
[3]
[3]
[4]
χ(s)
1/2
u(p, s) = (E + m) σ ·~
~ p
E+m
χ(s)
where
(1) 1 (2) 0
χ = , χ = .
0 1
[10]